Why most agencies optimise the wrong KPI — and the financial cost of misaligned performance metrics
The most common failure mode in performance marketing is not bad creative, wrong audiences, or weak landing pages. It is optimising toward a metric that does not reflect business health. Platform ROAS looks correct. CTR benchmarks are being hit. Impression share is growing. Revenue is growing too — but contribution margin is eroding, CAC is climbing, and the business is scaling toward a profitable-looking unprofitable outcome.
Executive summary
Agencies optimise toward the metrics they are measured against. If a performance agency is evaluated on platform-reported ROAS, it will optimise campaigns to maximise platform-reported ROAS. If a brand's success metric is cost-per-lead without a quality filter, the agency will optimise campaigns to minimise cost-per-lead — and lead quality will degrade systematically until the metric is changed. This is not agency failure. It is incentive alignment functioning exactly as designed.
The correct metrics for a performance marketing engagement are business outcomes, not platform metrics. Contribution margin (revenue minus cost of goods minus direct acquisition cost) not ROAS. Customer lifetime value divided by customer acquisition cost, not cost-per-click. Qualified pipeline value, not lead volume. New customer revenue versus returning customer revenue, not blended revenue. These metrics require integration between ad platform data, CRM data, and financial data — which is why most agencies do not measure them. They require access they do not typically have.
The consequence of KPI misalignment is that a campaign can be 'succeeding' on every reported metric while the business is moving in the wrong direction. Optimising toward platform ROAS on high-AOV products with low contribution margins produces campaigns that scale revenue while eroding profitability. Optimising toward cost-per-lead for B2B SaaS without a lead quality filter produces campaigns that scale form submissions while saturating the sales team with unqualified pipeline.
The real problem
The agency is optimising the campaign correctly. It is optimising toward the wrong outcome.
When a brand instructs an agency to 'optimise for ROAS,' the agency configures campaign objectives, bidding strategies, and audience targeting to maximise the ratio of reported revenue to ad spend — as defined by the ad platform's attribution model. The agency is doing exactly what it was asked to do. If the ad platform's attribution model over-credits Meta conversions (as described elsewhere in this library), if the ROAS metric includes returning customers' purchases that would have happened without the ad, or if the product margin structure means a 3× ROAS is actually unprofitable, the agency has no visibility into any of these factors — and no incentive to ask.
The structural problem is that agencies typically have access to ad platform data and analytics, but not to financial data. They can see that ROAS is 3.4×. They cannot see that the product in the ad has a 28% gross margin, making 3.4× reported ROAS equivalent to approximately 0.95× contribution margin ROAS — i.e., losing money on every acquisition. This is not a hypothetical. It is a common configuration in ecommerce brands that sell both high-margin and low-margin products and run unified campaigns without product-level margin segmentation.
The same structural disconnect exists in B2B. An agency optimising for cost-per-lead generates the lowest-cost leads available. Low-cost leads are typically low-intent leads — broad search terms, generic content downloads, non-specific ad targeting. The cost-per-lead metric looks excellent. The sales team receives high volume, low-quality pipeline. Conversion rate from lead to opportunity falls. Sales cycle length increases. The business concludes paid media 'doesn't work for B2B' when the correct diagnosis is KPI misalignment.
The diagnostic question: if your agency sent you a report tomorrow showing all campaigns are hitting ROAS targets, CTR benchmarks, and lead volume goals — but contribution margin has been eroding for three months — would you know? Would they know?
Strategic breakdown
Six metrics that sound correct but optimise toward the wrong outcome.
Platform ROAS without margin segmentation. A blended 3.4× ROAS across products with margins ranging from 15% to 65% is not one metric — it is a weighted average of radically different business outcomes. Campaigns optimising toward blended ROAS systematically shift spend toward high-AOV, low-margin products because they produce the highest nominal ROAS. The correct metric is contribution margin ROAS: (revenue × gross margin − ad spend) ÷ ad spend.
Cost-per-lead without quality filter. For any business with a sales or qualification step between lead and revenue, volume-based cost-per-lead is an anti-metric — it actively incentivises the acquisition of unqualified leads that waste sales capacity. The correct metric is cost-per-qualified-lead or cost-per-opportunity, measured at the CRM stage that represents genuine purchase intent. This requires CRM-to-ad-platform integration that most agencies do not set up.
Blended ROAS including returning customers. Retargeting campaigns that reach existing customers who would have purchased anyway produce high ROAS (because the organic purchase is credited to the ad) at the cost of retargeting budget that could be acquiring new customers. The correct metric separates new customer revenue from returning customer revenue. New customer acquisition cost is the only meaningful measure of whether paid media is actually growing the business.
Impressions and reach for brand campaigns. Brand awareness campaigns are commonly reported on impressions, reach, and frequency — none of which measure brand health, purchase intent, or eventual revenue impact. The correct evaluation is brand search volume uplift, direct traffic growth, and controlled brand lift studies — metrics that measure the actual outcome of brand investment.
CTR as a creative quality signal. Click-through rate measures whether an ad generates clicks — not whether it generates qualified clicks that convert. In GCC markets with high mobile penetration, accidental clicks from mobile browsing inflate CTR without corresponding conversion value. The correct creative quality signals are conversion rate (not just CTR) and, at scale, cost-per-verified-acquisition from server-side attributed data.
Cost-per-install for app campaigns. The mobile app equivalent of cost-per-lead — optimising toward install volume produces the cheapest-to-acquire users, who are systematically the least likely to activate and retain. The correct metric is cost-per-retained-user (users active at day-7 or day-30 post-install). This requires mobile measurement partner (MMP) integration and takes 6–8 weeks of data to establish as a reliable optimisation signal.
System-level insight
The metric you give an agency determines the business outcome you get.
Metric design is strategy. The performance metric an agency is evaluated against becomes the decision function for every campaign, creative, audience, and bid strategy choice the agency makes. A well-specified metric — one that directly reflects a business outcome the brand cares about — produces campaigns that move in the right direction even when individual decisions are imperfect. A poorly specified metric produces campaigns that hit their target while the business moves in the wrong direction.
The brands with the best performance marketing outcomes are not always the ones with the best agencies or the highest budgets. They are the ones with the clearest metric alignment between what is reported and what the business actually needs. They have built the data infrastructure to measure contribution margin by campaign, qualified lead volume by source, and new customer revenue separately from returning customer revenue. They have given their agency these metrics rather than platform defaults. And they have built reporting that makes the business health signal — not the platform health signal — the primary performance indicator.
Building this alignment requires investment in measurement infrastructure that goes beyond standard analytics and ad platform setup. CRM integration, back-end order data import, margin data feed by product — these are engineering decisions that determine whether the business can measure what matters. Brands that make this investment gain a compounding advantage: every optimisation cycle is driving toward a metric that reflects actual business health, not a proxy that diverges from it at scale.
Operational implications
If your agency is currently reporting on platform ROAS, CTR, and lead volume, these four diagnostics will tell you whether these metrics are producing the business outcomes you actually need.
Calculate contribution margin ROAS
Take your blended reported ROAS for last month. Multiply by your blended gross margin across the product mix in paid campaigns. If the result is below 1.0, you are losing money on every paid acquisition before accounting for overhead. If the result is below 1.5, your acquisition economics are marginal and any efficiency degradation produces losses. The reported ROAS may be 3.4×. The contribution margin ROAS may be 0.9×.
Separate new vs returning customer revenue from paid media
In your CRM or ecommerce platform, filter paid media-attributed orders by new customer vs returning customer. What percentage of paid media revenue is from customers who had not previously purchased? If below 60%, a significant portion of your paid media spend is subsidising purchases from existing customers who would have bought anyway. This is a retargeting allocation problem, not a creative problem.
Measure lead-to-qualified rate by source
For B2B, pull your CRM data: of all leads generated from paid media in the last 90 days, what percentage reached a defined qualified stage (e.g., Sales Accepted Lead, Opportunity Created)? Break this down by campaign and ad set. If your best-performing campaigns by CPL have the lowest lead-to-qualified rates, your optimisation metric is anti-correlated with your business outcome.
Request a contribution margin dashboard
Ask your agency to build a reporting view that shows campaign performance against contribution margin, not ROAS. If the agency cannot build this (because they do not have access to your margin data), that is the first problem to solve — not the campaign settings. Sharing product-level margin data with your performance agency is the prerequisite for margin-aligned campaign optimisation.
Recommended architecture
The business-aligned KPI framework.
This is the metric architecture we implement for performance engagements — replacing platform defaults with business-outcome metrics that remain reliable as spend scales. The framework applies across ecommerce, SaaS, and lead generation business models.
Contribution margin data feed
Connect product-level gross margin data to your ad platform reporting — via Shopify product metafields, a data warehouse integration, or a manual margin lookup table synced weekly. This enables campaign-level contribution margin ROAS calculation rather than revenue ROAS. For ecommerce: set campaign-level bid targets based on contribution margin per acquisition, not revenue ROAS.
New vs returning customer segmentation
Configure your CRM and ad platform reporting to distinguish new customer acquisitions from returning customer purchases. In Meta, use the 'New Customer Acquisition' campaign objective or segment attribution by customer email match against existing customer list. Report new customer CAC as the primary acquisition metric, with blended ROAS as a secondary indicator.
CRM-to-platform qualified lead import
For B2B and lead generation: configure CRM pipeline stage events as offline conversions imported to Meta and Google. Optimise campaigns toward qualified pipeline value (lead × average deal size × stage conversion rate), not lead volume. This changes the algorithm's optimisation target from form completion to business-valued pipeline — improving lead quality within 4–8 weeks of retraining.
LTV-adjusted CAC tracking
For subscription and SaaS businesses: calculate 12-month LTV for cohorts by acquisition channel and campaign. Apply the LTV:CAC ratio as the primary efficiency benchmark — not month-one ROAS. Set CAC targets derived from acceptable LTV payback periods (e.g., if 12-month LTV is AED 1,200 and target payback is 8 months, max acceptable CAC is AED 800).
Weekly business health review
Replace the standard weekly ROAS report with a weekly business health report: new customer revenue vs returning customer revenue, contribution margin by channel, qualified pipeline value by source (for B2B), and LTV:CAC by cohort (for subscription). Platform metrics remain visible but are not the primary decision inputs — they are diagnostic tools for explaining movement in the business health metrics.
Related services
From intelligence to system
The architecture described above is available as an engagement.
We start with a diagnostic — identifying the specific layer that is constraining your current growth. No generic proposals. No long retainers before results are visible.
- Senior strategist on every engagement
- UAE · KSA · Global markets
- Diagnostic-first, not deck-first